Deepseek has hit the table this week, causing billions of losses in many companies, especially in Nvidia who lost more than 600,000 million dollars overnight in stock market capitalization making it lowering it from the first position it led to the third and be overcome by Apple and Microsoft.
The main and only reason for what some consider the end of the AI bubbleis the cost of having developed this model. According to Depseek the company has only needed 6 million dollars To create it.
OpenAi’s model has had a cost of 100 million dollars. The difference between the two is abysmal and how could it be otherwise, the United States, like other companies, doubts a lot about the veracity of that data.
What GPUS has been used to create Depseek R1?
The cost of your training is linked to the hardware that has been used. According to the company, Deepseek R1 has been trained with the H800 GPUa graph of Nvidia launched to the market in 2023 and that was sold without limitations to China until the restrictions of the administration of Joe Biden in 2024 arrived.
Based on this information, the market wondered why pay more for the NVIDIA H100 if we can obtain the same result with the H800 of 2023?
If the H800 allows to obtain the same result as the H100, GPU much more expensive than the H800, it makes no sense to continue betting on the latest generation of NVIDIA GPUS. This was, and not another, the main reason for the fall in the value of the actions we have commented at the beginning of this article.
The United States does not trust
The reason to doubt this figure is due to the proper functioning of Deepseek R1, very similar to the one OpenAi offers with GPT 01 Able to think about your answers, that is, look for the correct option to the question contrasting the information with a greater number of sources in order to offer the correct answer.
To give light to this matter, according to Bloomberg, Donald Trump’s government has opened an investigation to find out if Depseek has used NVIDIA H-100 GPUs for training, GPUS that Nvidia is forbidden to sell to China precisely to prevent me It will advance as fast as it has done in the field of artificial intelligence.
The United States Government wants to determine if the company has used intermediary to buy NVIDIA H100 GPU, thus avoiding the United States commercial policies. According to Howard Lutnick, from the United States Department of Commerce:
This has to end. If they are going to compete with us, who compete, but stop using our tools to compete with us.
One of the countries that is among the suspects is Singapore, a country that represents the 20% of Nvidia sales. To prevent China from using intermediary countries, it has expanded the number of countries with which Nvidia cannot trade more than 40, but not including in the Singapore list. This is because many companies have offices in this country that they use for products that are sent from China to the rest of the world.
